Approaching homeostasis in a binary neural network

A system comprising:

• a computing node comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor of the computing node to cause the processor to perform a method comprising:
• initializing a region of an artificial neural network,
• the region comprising a plurality of neurons and having a permanence value associated with each potential synaptic connection between neurons,
• said initializing comprising connecting a subset of the potential synaptic connections by synapses;
• receiving a plurality of time-ordered inputs to the region, thereby causing some of the plurality of neurons to fire upon receipt of each time-ordered input;
• and
• upon receipt of each time-ordered input, for each potential synaptic connection between neurons,
• (i) adjusting the permanence value according to a firing sequence of the plurality of neurons;
• (ii) connecting those potential synaptic connections having a permanence value above a predetermined permanence threshold when the total number of connected synapses in the region does not exceed a predetermined connectivity threshold;
• and
• (iii) disconnecting synapses whose associated pre-synaptic neuron and/or post-synaptic neuron have more than a predetermined number of connected synapses.

A computer program product for operating a neural network, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to perform a method comprising:

• initializing a region of an artificial neural network,
• the region comprising a plurality of neurons and having a permanence value associated with each potential synaptic connection between neurons,
• said initializing comprising connecting a subset of the potential synaptic connections by synapses;
• receiving a plurality of time-ordered inputs to the region, thereby causing some of the plurality of neurons to fire upon receipt of each time-ordered input;
• and
• upon receipt of each time-ordered input, for each potential synaptic connection between neurons:
• (i) adjusting the permanence value according to a firing sequence of the plurality of neurons;
• (ii) connecting those potential synaptic connection having a permanence value above a predetermined permanence threshold when the total number of connected synapses in the region does not exceed a predetermined connectivity threshold;
• and
• (iii) disconnecting synapses whose associated pre-synaptic neuron and/or post-synaptic neuron have more than a predetermined number of connected synapses.

An artificial neural network comprising a region comprising a plurality of neurons and having a permanence value associated with each potential synaptic connection between neurons, the neural network adapted to:

• initialize the region of an artificial neural network by connecting a subset of the potential synaptic connections by synapses;
• receive a plurality of time-ordered inputs to the region, thereby causing some of the plurality of neurons to fire upon receipt of each time-ordered input;
• and
• upon receipt of each time-ordered input, for each potential synaptic connection between neurons,
• (i) adjust the permanence value according to a firing sequence of the plurality of neurons;
• (ii) connect those potential synaptic connections having a permanence value above a predetermined permanence threshold when the total number of connected synapses in the region does not exceed a predetermined connectivity threshold;
• and
• (iii) disconnect synapses whose associated pre-synaptic neuron and/or post-synaptic neuron have more than a predetermined number of connected synapses.

A method comprising:

• receiving a plurality of time-ordered inputs to an artificial neural network, thereby causing a plurality of neurons of the artificial neural network to fire upon receipt of each time-ordered input;
• upon receipt of each time-ordered input, for each potential synaptic connection between neurons of the artificial neural network:
• (i) increasing a permanence value associated with each potential synaptic connection between neurons that fire in sequence and (ii) decreasing a permanence value associated with each potential synaptic connection between neurons that do not fire in sequence;
• connecting those potential synaptic connections having a permanence value above a predetermined permanence threshold when the total number of connected synapses does not exceed a predetermined connectivity threshold;
• and
• disconnecting synaptic connections whose associated neurons have more than a predetermined number of synaptic connections.